/* * pSeries NUMA support * * Copyright (C) 2002 Anton Blanchard , IBM * * This program is free software; you can redistribute it and/or * modify it under the terms of the GNU General Public License * as published by the Free Software Foundation; either version * 2 of the License, or (at your option) any later version. */ #include #include #include #include #include #include #include #include #include #include #include #include #include static int numa_enabled = 1; static int numa_debug; #define dbg(args...) if (numa_debug) { printk(KERN_INFO args); } int numa_cpu_lookup_table[NR_CPUS]; cpumask_t numa_cpumask_lookup_table[MAX_NUMNODES]; struct pglist_data *node_data[MAX_NUMNODES]; EXPORT_SYMBOL(numa_cpu_lookup_table); EXPORT_SYMBOL(numa_cpumask_lookup_table); EXPORT_SYMBOL(node_data); static bootmem_data_t __initdata plat_node_bdata[MAX_NUMNODES]; static int min_common_depth; static int n_mem_addr_cells, n_mem_size_cells; /* * We need somewhere to store start/end/node for each region until we have * allocated the real node_data structures. */ #define MAX_REGIONS (MAX_LMB_REGIONS*2) static struct { unsigned long start_pfn; unsigned long end_pfn; int nid; } init_node_data[MAX_REGIONS] __initdata; int __init early_pfn_to_nid(unsigned long pfn) { unsigned int i; for (i = 0; init_node_data[i].end_pfn; i++) { unsigned long start_pfn = init_node_data[i].start_pfn; unsigned long end_pfn = init_node_data[i].end_pfn; if ((start_pfn <= pfn) && (pfn < end_pfn)) return init_node_data[i].nid; } return -1; } void __init add_region(unsigned int nid, unsigned long start_pfn, unsigned long pages) { unsigned int i; dbg("add_region nid %d start_pfn 0x%lx pages 0x%lx\n", nid, start_pfn, pages); for (i = 0; init_node_data[i].end_pfn; i++) { if (init_node_data[i].nid != nid) continue; if (init_node_data[i].end_pfn == start_pfn) { init_node_data[i].end_pfn += pages; return; } if (init_node_data[i].start_pfn == (start_pfn + pages)) { init_node_data[i].start_pfn -= pages; return; } } /* * Leave last entry NULL so we dont iterate off the end (we use * entry.end_pfn to terminate the walk). */ if (i >= (MAX_REGIONS - 1)) { printk(KERN_ERR "WARNING: too many memory regions in " "numa code, truncating\n"); return; } init_node_data[i].start_pfn = start_pfn; init_node_data[i].end_pfn = start_pfn + pages; init_node_data[i].nid = nid; } /* We assume init_node_data has no overlapping regions */ void __init get_region(unsigned int nid, unsigned long *start_pfn, unsigned long *end_pfn, unsigned long *pages_present) { unsigned int i; *start_pfn = -1UL; *end_pfn = *pages_present = 0; for (i = 0; init_node_data[i].end_pfn; i++) { if (init_node_data[i].nid != nid) continue; *pages_present += init_node_data[i].end_pfn - init_node_data[i].start_pfn; if (init_node_data[i].start_pfn < *start_pfn) *start_pfn = init_node_data[i].start_pfn; if (init_node_data[i].end_pfn > *end_pfn) *end_pfn = init_node_data[i].end_pfn; } /* We didnt find a matching region, return start/end as 0 */ if (*start_pfn == -1UL) *start_pfn = 0; } static inline void map_cpu_to_node(int cpu, int node) { numa_cpu_lookup_table[cpu] = node; if (!(cpu_isset(cpu, numa_cpumask_lookup_table[node]))) cpu_set(cpu, numa_cpumask_lookup_table[node]); } #ifdef CONFIG_HOTPLUG_CPU static void unmap_cpu_from_node(unsigned long cpu) { int node = numa_cpu_lookup_table[cpu]; dbg("removing cpu %lu from node %d\n", cpu, node); if (cpu_isset(cpu, numa_cpumask_lookup_table[node])) { cpu_clear(cpu, numa_cpumask_lookup_table[node]); } else { printk(KERN_ERR "WARNING: cpu %lu not found in node %d\n", cpu, node); } } #endif /* CONFIG_HOTPLUG_CPU */ static struct device_node *find_cpu_node(unsigned int cpu) { unsigned int hw_cpuid = get_hard_smp_processor_id(cpu); struct device_node *cpu_node = NULL; unsigned int *interrupt_server, *reg; int len; while ((cpu_node = of_find_node_by_type(cpu_node, "cpu")) != NULL) { /* Try interrupt server first */ interrupt_server = (unsigned int *)get_property(cpu_node, "ibm,ppc-interrupt-server#s", &len); len = len / sizeof(u32); if (interrupt_server && (len > 0)) { while (len--) { if (interrupt_server[len] == hw_cpuid) return cpu_node; } } else { reg = (unsigned int *)get_property(cpu_node, "reg", &len); if (reg && (len > 0) && (reg[0] == hw_cpuid)) return cpu_node; } } return NULL; } /* must hold reference to node during call */ static int *of_get_associativity(struct device_node *dev) { return (unsigned int *)get_property(dev, "ibm,associativity", NULL); } static int of_node_numa_domain(struct device_node *device) { int numa_domain; unsigned int *tmp; if (min_common_depth == -1) return 0; tmp = of_get_associativity(device); if (tmp && (tmp[0] >= min_common_depth)) { numa_domain = tmp[min_common_depth]; } else { dbg("WARNING: no NUMA information for %s\n", device->full_name); numa_domain = 0; } return numa_domain; } /* * In theory, the "ibm,associativity" property may contain multiple * associativity lists because a resource may be multiply connected * into the machine. This resource then has different associativity * characteristics relative to its multiple connections. We ignore * this for now. We also assume that all cpu and memory sets have * their distances represented at a common level. This won't be * true for heirarchical NUMA. * * In any case the ibm,associativity-reference-points should give * the correct depth for a normal NUMA system. * * - Dave Hansen */ static int __init find_min_common_depth(void) { int depth; unsigned int *ref_points; struct device_node *rtas_root; unsigned int len; rtas_root = of_find_node_by_path("/rtas"); if (!rtas_root) return -1; /* * this property is 2 32-bit integers, each representing a level of * depth in the associativity nodes. The first is for an SMP * configuration (should be all 0's) and the second is for a normal * NUMA configuration. */ ref_points = (unsigned int *)get_property(rtas_root, "ibm,associativity-reference-points", &len); if ((len >= 1) && ref_points) { depth = ref_points[1]; } else { dbg("WARNING: could not find NUMA " "associativity reference point\n"); depth = -1; } of_node_put(rtas_root); return depth; } static void __init get_n_mem_cells(int *n_addr_cells, int *n_size_cells) { struct device_node *memory = NULL; memory = of_find_node_by_type(memory, "memory"); if (!memory) panic("numa.c: No memory nodes found!"); *n_addr_cells = prom_n_addr_cells(memory); *n_size_cells = prom_n_size_cells(memory); of_node_put(memory); } static unsigned long __devinit read_n_cells(int n, unsigned int **buf) { unsigned long result = 0; while (n--) { result = (result << 32) | **buf; (*buf)++; } return result; } /* * Figure out to which domain a cpu belongs and stick it there. * Return the id of the domain used. */ static int numa_setup_cpu(unsigned long lcpu) { int numa_domain = 0; struct device_node *cpu = find_cpu_node(lcpu); if (!cpu) { WARN_ON(1); goto out; } numa_domain = of_node_numa_domain(cpu); if (numa_domain >= num_online_nodes()) { /* * POWER4 LPAR uses 0xffff as invalid node, * dont warn in this case. */ if (numa_domain != 0xffff) printk(KERN_ERR "WARNING: cpu %ld " "maps to invalid NUMA node %d\n", lcpu, numa_domain); numa_domain = 0; } out: node_set_online(numa_domain); map_cpu_to_node(lcpu, numa_domain); of_node_put(cpu); return numa_domain; } static int cpu_numa_callback(struct notifier_block *nfb, unsigned long action, void *hcpu) { unsigned long lcpu = (unsigned long)hcpu; int ret = NOTIFY_DONE; switch (action) { case CPU_UP_PREPARE: if (min_common_depth == -1 || !numa_enabled) map_cpu_to_node(lcpu, 0); else numa_setup_cpu(lcpu); ret = NOTIFY_OK; break; #ifdef CONFIG_HOTPLUG_CPU case CPU_DEAD: case CPU_UP_CANCELED: unmap_cpu_from_node(lcpu); break; ret = NOTIFY_OK; #endif } return ret; } /* * Check and possibly modify a memory region to enforce the memory limit. * * Returns the size the region should have to enforce the memory limit. * This will either be the original value of size, a truncated value, * or zero. If the returned value of size is 0 the region should be * discarded as it lies wholy above the memory limit. */ static unsigned long __init numa_enforce_memory_limit(unsigned long start, unsigned long size) { /* * We use lmb_end_of_DRAM() in here instead of memory_limit because * we've already adjusted it for the limit and it takes care of * having memory holes below the limit. */ if (! memory_limit) return size; if (start + size <= lmb_end_of_DRAM()) return size; if (start >= lmb_end_of_DRAM()) return 0; return lmb_end_of_DRAM() - start; } static int __init parse_numa_properties(void) { struct device_node *cpu = NULL; struct device_node *memory = NULL; int max_domain; unsigned long i; if (numa_enabled == 0) { printk(KERN_WARNING "NUMA disabled by user\n"); return -1; } min_common_depth = find_min_common_depth(); dbg("NUMA associativity depth for CPU/Memory: %d\n", min_common_depth); if (min_common_depth < 0) return min_common_depth; max_domain = numa_setup_cpu(boot_cpuid); /* * Even though we connect cpus to numa domains later in SMP init, * we need to know the maximum node id now. This is because each * node id must have NODE_DATA etc backing it. * As a result of hotplug we could still have cpus appear later on * with larger node ids. In that case we force the cpu into node 0. */ for_each_cpu(i) { int numa_domain; cpu = find_cpu_node(i); if (cpu) { numa_domain = of_node_numa_domain(cpu); of_node_put(cpu); if (numa_domain < MAX_NUMNODES && max_domain < numa_domain) max_domain = numa_domain; } } get_n_mem_cells(&n_mem_addr_cells, &n_mem_size_cells); memory = NULL; while ((memory = of_find_node_by_type(memory, "memory")) != NULL) { unsigned long start; unsigned long size; int numa_domain; int ranges; unsigned int *memcell_buf; unsigned int len; memcell_buf = (unsigned int *)get_property(memory, "linux,usable-memory", &len); if (!memcell_buf || len <= 0) memcell_buf = (unsigned int *)get_property(memory, "reg", &len); if (!memcell_buf || len <= 0) continue; ranges = memory->n_addrs; new_range: /* these are order-sensitive, and modify the buffer pointer */ start = read_n_cells(n_mem_addr_cells, &memcell_buf); size = read_n_cells(n_mem_size_cells, &memcell_buf); numa_domain = of_node_numa_domain(memory); if (numa_domain >= MAX_NUMNODES) { if (numa_domain != 0xffff) printk(KERN_ERR "WARNING: memory at %lx maps " "to invalid NUMA node %d\n", start, numa_domain); numa_domain = 0; } if (max_domain < numa_domain) max_domain = numa_domain; if (!(size = numa_enforce_memory_limit(start, size))) { if (--ranges) goto new_range; else continue; } add_region(numa_domain, start >> PAGE_SHIFT, size >> PAGE_SHIFT); if (--ranges) goto new_range; } for (i = 0; i <= max_domain; i++) node_set_online(i); return 0; } static void __init setup_nonnuma(void) { unsigned long top_of_ram = lmb_end_of_DRAM(); unsigned long total_ram = lmb_phys_mem_size(); unsigned int i; printk(KERN_INFO "Top of RAM: 0x%lx, Total RAM: 0x%lx\n", top_of_ram, total_ram); printk(KERN_INFO "Memory hole size: %ldMB\n", (top_of_ram - total_ram) >> 20); map_cpu_to_node(boot_cpuid, 0); for (i = 0; i < lmb.memory.cnt; ++i) add_region(0, lmb.memory.region[i].base >> PAGE_SHIFT, lmb_size_pages(&lmb.memory, i)); node_set_online(0); } void __init dump_numa_cpu_topology(void) { unsigned int node; unsigned int cpu, count; if (min_common_depth == -1 || !numa_enabled) return; for_each_online_node(node) { printk(KERN_INFO "Node %d CPUs:", node); count = 0; /* * If we used a CPU iterator here we would miss printing * the holes in the cpumap. */ for (cpu = 0; cpu < NR_CPUS; cpu++) { if (cpu_isset(cpu, numa_cpumask_lookup_table[node])) { if (count == 0) printk(" %u", cpu); ++count; } else { if (count > 1) printk("-%u", cpu - 1); count = 0; } } if (count > 1) printk("-%u", NR_CPUS - 1); printk("\n"); } } static void __init dump_numa_memory_topology(void) { unsigned int node; unsigned int count; if (min_common_depth == -1 || !numa_enabled) return; for_each_online_node(node) { unsigned long i; printk(KERN_INFO "Node %d Memory:", node); count = 0; for (i = 0; i < lmb_end_of_DRAM(); i += (1 << SECTION_SIZE_BITS)) { if (early_pfn_to_nid(i >> PAGE_SHIFT) == node) { if (count == 0) printk(" 0x%lx", i); ++count; } else { if (count > 0) printk("-0x%lx", i); count = 0; } } if (count > 0) printk("-0x%lx", i); printk("\n"); } } /* * Allocate some memory, satisfying the lmb or bootmem allocator where * required. nid is the preferred node and end is the physical address of * the highest address in the node. * * Returns the physical address of the memory. */ static void __init *careful_allocation(int nid, unsigned long size, unsigned long align, unsigned long end_pfn) { int new_nid; unsigned long ret = lmb_alloc_base(size, align, end_pfn << PAGE_SHIFT); /* retry over all memory */ if (!ret) ret = lmb_alloc_base(size, align, lmb_end_of_DRAM()); if (!ret) panic("numa.c: cannot allocate %lu bytes on node %d", size, nid); /* * If the memory came from a previously allocated node, we must * retry with the bootmem allocator. */ new_nid = early_pfn_to_nid(ret >> PAGE_SHIFT); if (new_nid < nid) { ret = (unsigned long)__alloc_bootmem_node(NODE_DATA(new_nid), size, align, 0); if (!ret) panic("numa.c: cannot allocate %lu bytes on node %d", size, new_nid); ret = __pa(ret); dbg("alloc_bootmem %lx %lx\n", ret, size); } return (void *)ret; } void __init do_init_bootmem(void) { int nid; unsigned int i; static struct notifier_block ppc64_numa_nb = { .notifier_call = cpu_numa_callback, .priority = 1 /* Must run before sched domains notifier. */ }; min_low_pfn = 0; max_low_pfn = lmb_end_of_DRAM() >> PAGE_SHIFT; max_pfn = max_low_pfn; if (parse_numa_properties()) setup_nonnuma(); else dump_numa_memory_topology(); register_cpu_notifier(&ppc64_numa_nb); for_each_online_node(nid) { unsigned long start_pfn, end_pfn, pages_present; unsigned long bootmem_paddr; unsigned long bootmap_pages; get_region(nid, &start_pfn, &end_pfn, &pages_present); /* Allocate the node structure node local if possible */ NODE_DATA(nid) = careful_allocation(nid, sizeof(struct pglist_data), SMP_CACHE_BYTES, end_pfn); NODE_DATA(nid) = __va(NODE_DATA(nid)); memset(NODE_DATA(nid), 0, sizeof(struct pglist_data)); dbg("node %d\n", nid); dbg("NODE_DATA() = %p\n", NODE_DATA(nid)); NODE_DATA(nid)->bdata = &plat_node_bdata[nid]; NODE_DATA(nid)->node_start_pfn = start_pfn; NODE_DATA(nid)->node_spanned_pages = end_pfn - start_pfn; if (NODE_DATA(nid)->node_spanned_pages == 0) continue; dbg("start_paddr = %lx\n", start_pfn << PAGE_SHIFT); dbg("end_paddr = %lx\n", end_pfn << PAGE_SHIFT); bootmap_pages = bootmem_bootmap_pages(end_pfn - start_pfn); bootmem_paddr = (unsigned long)careful_allocation(nid, bootmap_pages << PAGE_SHIFT, PAGE_SIZE, end_pfn); memset(__va(bootmem_paddr), 0, bootmap_pages << PAGE_SHIFT); dbg("bootmap_paddr = %lx\n", bootmem_paddr); init_bootmem_node(NODE_DATA(nid), bootmem_paddr >> PAGE_SHIFT, start_pfn, end_pfn); /* Add free regions on this node */ for (i = 0; init_node_data[i].end_pfn; i++) { unsigned long start, end; if (init_node_data[i].nid != nid) continue; start = init_node_data[i].start_pfn << PAGE_SHIFT; end = init_node_data[i].end_pfn << PAGE_SHIFT; dbg("free_bootmem %lx %lx\n", start, end - start); free_bootmem_node(NODE_DATA(nid), start, end - start); } /* Mark reserved regions on this node */ for (i = 0; i < lmb.reserved.cnt; i++) { unsigned long physbase = lmb.reserved.region[i].base; unsigned long size = lmb.reserved.region[i].size; unsigned long start_paddr = start_pfn << PAGE_SHIFT; unsigned long end_paddr = end_pfn << PAGE_SHIFT; if (early_pfn_to_nid(physbase >> PAGE_SHIFT) != nid && early_pfn_to_nid((physbase+size-1) >> PAGE_SHIFT) != nid) continue; if (physbase < end_paddr && (physbase+size) > start_paddr) { /* overlaps */ if (physbase < start_paddr) { size -= start_paddr - physbase; physbase = start_paddr; } if (size > end_paddr - physbase) size = end_paddr - physbase; dbg("reserve_bootmem %lx %lx\n", physbase, size); reserve_bootmem_node(NODE_DATA(nid), physbase, size); } } /* Add regions into sparsemem */ for (i = 0; init_node_data[i].end_pfn; i++) { unsigned long start, end; if (init_node_data[i].nid != nid) continue; start = init_node_data[i].start_pfn; end = init_node_data[i].end_pfn; memory_present(nid, start, end); } } } void __init paging_init(void) { unsigned long zones_size[MAX_NR_ZONES]; unsigned long zholes_size[MAX_NR_ZONES]; int nid; memset(zones_size, 0, sizeof(zones_size)); memset(zholes_size, 0, sizeof(zholes_size)); for_each_online_node(nid) { unsigned long start_pfn, end_pfn, pages_present; get_region(nid, &start_pfn, &end_pfn, &pages_present); zones_size[ZONE_DMA] = end_pfn - start_pfn; zholes_size[ZONE_DMA] = zones_size[ZONE_DMA] - pages_present; dbg("free_area_init node %d %lx %lx (hole: %lx)\n", nid, zones_size[ZONE_DMA], start_pfn, zholes_size[ZONE_DMA]); free_area_init_node(nid, NODE_DATA(nid), zones_size, start_pfn, zholes_size); } } static int __init early_numa(char *p) { if (!p) return 0; if (strstr(p, "off")) numa_enabled = 0; if (strstr(p, "debug")) numa_debug = 1; return 0; } early_param("numa", early_numa); #ifdef CONFIG_MEMORY_HOTPLUG /* * Find the node associated with a hot added memory section. Section * corresponds to a SPARSEMEM section, not an LMB. It is assumed that * sections are fully contained within a single LMB. */ int hot_add_scn_to_nid(unsigned long scn_addr) { struct device_node *memory = NULL; if (!numa_enabled || (min_common_depth < 0)) return 0; while ((memory = of_find_node_by_type(memory, "memory")) != NULL) { unsigned long start, size; int numa_domain, ranges; unsigned int *memcell_buf; unsigned int len; memcell_buf = (unsigned int *)get_property(memory, "reg", &len); if (!memcell_buf || len <= 0) continue; ranges = memory->n_addrs; /* ranges in cell */ ha_new_range: start = read_n_cells(n_mem_addr_cells, &memcell_buf); size = read_n_cells(n_mem_size_cells, &memcell_buf); numa_domain = of_node_numa_domain(memory); /* Domains not present at boot default to 0 */ if (!node_online(numa_domain)) numa_domain = any_online_node(NODE_MASK_ALL); if ((scn_addr >= start) && (scn_addr < (start + size))) { of_node_put(memory); return numa_domain; } if (--ranges) /* process all ranges in cell */ goto ha_new_range; } BUG(); /* section address should be found above */ return 0; } #endif /* CONFIG_MEMORY_HOTPLUG */